Recent technological advances have made it possible to move a robotic device in real time, using signals obtained directly from the brain. This field of Brain Machine Interface (BMI) has the means to provide movement for paralyzed patients, communication for locked-in patients, and a better understanding of the brain for all of society. In order to control movement effectively, the brain must be able to activate muscles appropriately, and monitor the evolving movement quickly and precisely. Existing BMIs for, while remarkable, do each of these tasks in poor imitation of the intact nervous system. Our proposed work addresses these limitations by developing a bidirectional interface that produces movement in a more natural way, and provides feedback about the movement by direct, electrical stimulation of the brain. Our partnership includes members at Northwestern Univ (NU), Univ of Chicago (UC), Univ of Mass, Amherst (UMass), and the autonomous Univ of Mexico (UNAM). Partners have advanced degrees in a range of biological science, computer science, physics, mathematics, and engineering disciplines. Miller (NU) will coordinate the partnership. He has extensive experience with a wide range of recording, stimulation and behavioral protocols in behaving monkeys. Hatsopoulos (UC) is at the forefront of the field of multi-electrode recordings. He was a leading member of the first group to demonstrate visually guided BMI control by a primate. Barto (U Mass) has done pioneering research in neural networks, machine learning and stochastic optimization. Fagg (UMass) is an authority in the control of reaching and grasping robots that learn to interact with the environment. Together they will develop the decoders of activity from the brain used to cause movement. Romo (UNAM), is a world leader in studies of the perceptual and decision making processes induced by electrical stimulation of the brain. Solla (NU) is an expert in neural networks and information theory. With Romo, she will develop optimal routines to encode information in stimulus trains to provide feedback to the brain. Mussa-Ivaldi (NU) will focus on the overall design and evaluation of the interfaces. He created the first ever bidirectional interface between neural tissue and a robotic device.